- This topic has 3 replies, 2 voices, and was last updated 4 days, 7 hours ago by .
Viewing 4 posts - 1 through 4 (of 4 total)
Viewing 4 posts - 1 through 4 (of 4 total)
- You must be logged in to reply to this topic.
Home › Forums › FPGAs in FABRIC › STAR probably needs reflashing to P4 workflow
Hi FABRIC team,
We are having trouble standing up the EJFAT ESnet P4-based stack on STAR U280 and we suspect maybe it is not somehow flashed properly for the P4 workflow.
I’ve tried to use WASH as an alternative, but as I reported that has another problem (message).
We also have LOSA assigned to our project, but FPGA there is taken by someone (not from our project).
I’m out of options to rule out the flashing issue using another site as I don’t know which other site I can try, what I can say is this same notebook with exact same software and docker images was successfully tested right before Xmas on KANS and LOSA.
We are narrowing it down to PCI (or PCI passthrough) issue. From inside the VM we see this:
ubuntu@LB-node:~/esnet-smartnic-fw/sn-stack$ sudo lspci -Dd 10ee: -vv
0000:1f:00.0 Network controller: Xilinx Corporation Device 903f
Subsystem: Xilinx Corporation Device 0007
Physical Slot: 0-30
Control: I/O- Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
….
In the Control: line it says BusMaster- when it should instead be BusMaster+ (not sure what is visible from the server side)
Perhaps a cold-reboot of the server is needed (?). Not sure.
Hi Ilya,
I checked the host and see a similar output. Will check with Mert regarding the reboot.
0000:25:00.0 Network controller: Xilinx Corporation Device 903f
Subsystem: Xilinx Corporation Device 0007
Physical Slot: 2-1
Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx-
Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort-
Thanks,
Komal
Just to add – I stood up the same slice on WASH with no problems. It does also show BusMaster-
so this may have been a red herring.